An Younggil’s Pro Go Videos: DeepMind AlphaGo vs Lee Sedol – Game 4

This is a commentary of game four from the Google DeepMind Challenge Match.

It was played between AlphaGo and Lee Sedol 9p on March 13, 2016 in Seoul, Korea.

After Lee lost the first three games, his chance of winning the five game match had evaporated.

His revised goal, and the hope of millions of his fans, was that he might succeed in winning at least one game against the machine before the match concluded.

AlphaGo vs Lee Sedol – Game 4 of 5

Game record with variations


Download SGF File (Go Game Record)


Lee Sedol

Lee Sedol has won 14 world titles in his career, and was challenged by DeepMind because he has been the best Go player in the world over the last decade.

DeepMind AlphaGo

AlphaGo is an artificial intelligence which was developed by Google DeepMind.

In a previous match, AlphaGo defeated Fan Hui 2p 5–0, in October, 2015.

This is an historic match in Go, and it’s the first time that a computer program has challenged a human Go champion in an even game.

This video was recorded at the Sydney Go club in Sydney, Australia.

Related Articles

About Younggil An

Younggil is an 8 dan professional Go player with the Korean Baduk Association. He qualified as a professional in 1997 and won an award for winning 18 consecutive professional matches the following year. After completing compulsory military service, Younggil left Korea in 2008, to teach and promote the game Go overseas. Younggil now lives in Sydney, Australia, and is one of the founders of Go Game Guru. On Friday evenings, Younggil is usually at the Sydney Go Club, where he gives weekly lessons and plays simultaneous games.

You can follow Go Game Guru on Facebook, Twitter, Google+ and Youtube.


  1. It looks like Lee favor to

  2. cuong tran says:

    I don’t know why but it feel like Lee Sedol wasn’t try his best all 5 games.Black S8 then White S5 how can black live ?

  3. cuong tran says:

    Why Black didnt play L10 , J10 then G10 ??? I couldnt find any good move for white after that

  4. it looks like Lee was favor to play white, if he play black…
    1. if Lee play the 3rd move to make an enclose at the corner to secure some territory, the feeling of the game will be the same as he play white or not?
    2. if Lee play the first move at the middle and continue with mirrored game and use human judgment when to break the mirror, it will be better?

  5. 3. how about 101 x 101 board with Alphago? haha..

  6. cuong tran says:

    This is a good training machine .Korean Baduk should buy it .

  7. cuong tran says:

    If i play with Aphago everyday i can become a pro :3

  8. Humans vs Humans and computers vs computers please! 100 years ago there were matches between humans and cars, horses and cars… this is considered absurd now. With this match the beauty of Go was lost forever. Like chess after Deep Blue won over Garry Kasparov. Sad end.

  9. Could black have ignored 76 to play a move in the center? He would lose something at the top, but wouldn’t he still be ahead? I don’t see the corner dying.

  10. Wow GGG! Thank you for uploading another video and sgf with variations about this game! I’m looking forward to seeing your commentary, Younggil An. This is absolutely one of the most unforgettable games I ever watched and I will always look back on it fondly with LSD’s fighting spirit. You know, there’s a photo of him lurking around with a satisfied smile on his face after game 4 haha. 나는 그의 팬이에요! 이기거나 지거나, 이세돌은 나의 마음과 존경을 사로잡았다.

  11. 18 world titles … he has won I thinl,right?

  12. yes, 18 world titles (second in place behind 8 years older Lee Chang-ho who has won 21 titles). More about Lee Sedol (Yi Se-tol) and deep learning program AlphaGo you can find at

  13. sedolgogogo says:
    Lee sedol reviewed all five games. English translation’s here which is very impressive!

  14. Vladimir says:

    GoGameGuru, I know you folks are busy and all, but zero new articles after more than a month … 1 article since the AlphaGo vs Lee Sedol match?? Way to capitalize on the huge influx of new players … dayum!

  15. Was thinking the same thing Vladimir. Rather unfortunate state of affairs.

  16. There is a large amount of preexisting material for newbies to ponder.

    Think of it like this: busy = more sales = more future support for this and similar websites.

  17. I think the game can be brought into balance by having the timing changed . That is that the time of existence be set at a finite term . One way would be to set the existence of each stone to last some interval say 19 minutes per stone played . When the time is up that stone would vanish and the situation on the board would change thus equally handicapping the algorithm as the human . the advantage that the machine has is in counting moves plus prescribed patterns . One could even introduce an element of chance with Probability intervals . This way the machine would never know the count exactly .

  18. Anonymous says:

    hi younkil this is Gary how are u